Autonomous Pedestrian Collision Avoidance Using Fuzzy Steering Control

This paper presents the autonomous pedestrian collision avoidance using fuzzy steering control. There are three major components used to perform the pedestrian collision avoidance. First, a pedestrian detection using computer vision of TensorFlow SSDlite MonileNet v2, is developed to demonstrate the accuracy of pedestrian detection. Secondly, the Arduino board integrated with Fuzzy Logic System is developed to perform decision-making. Third one is the the fuzzy steering control in which two motors used as steering a brake paddle to perform the actuation. The performance of the developed system is evaluated by testing the Average Precision (AP) of the pedestrian detection, the speed of the pedestrian detection, the accuracy of ultrasonic sensor, the accuracy of speed sensor and the accuracy of Fuzzy Control System. The results are observed as 87% of accuracy on pedestrian detection and 99.97% of accuracy on determining the distance and 88% of accuracy on rpm determination.